Ellipsis is not Spell-out
What Scandinavian tells us about ellipsis and phases
Nicholas LaCara · University of Massachusetts Amherst
1
Introduction
Many languages, such as Portuguese (and Irish, and Hebrew, and Russian) typically
show verb stranding verb phrase ellipsis (vvpe):
(1)
Quando a Ana pôs
os óculos na
mesa, a Maria também
When the Ana put.pst the glasses on.the table, the Maria too
pôs
Δ.
put.pst
‘When Ana put the glasses on the table, the Maria did too.’
a
Portuguese (Cyrino and Matos 2002:(14a))
Following Goldberg (2005), the typical analysis is that a verb moves to a position
outside of the ellipsis site.
• This is the interaction of verb movement and verb phrase ellipsis (vpe).
(2)
Grasping Ellipsis · 5 May 2015
2. The Phase-based Approach:
Phase-based approaches propose that ellipsis and cyclic spell-out are directly
linked. Freezing effects are derived by the phase impenetrability condition.
I will defend the derivational approach as articulated by Sailor (2014).
• Aelbrecht’s (2010) derivational account of ellipsis actually predicts which languages will have vvpe without any modification.
• Regardless of whether ellipsis targets phase-head complements or whole
phases, phase-based approaches need additional stipulations to get the facts
straight.
[TP a Maria também pôs [VP por os óculos na mesa] ]
Roadmap:
A surprising fact is that Mainland Scandinavian languages do not have vvpe.
§2
How to strand a verb
Why the Goldberg view to verb stranding predicts Scandinavian
should have verb stranding.
§3
Derivational ellipsis
The derivational approach to ellipsis, and how it accounts for the lack
of stranding in Scandinavian but permits it in others.
• Verbs move out of νP to C° in matrix clauses, and it has vpe.
(3)
Mona vaskede ikke bilen men Jasper gjorde / *vaskede Δ.
Mona wash.pst not car.def, but Jasper do.pst / wash.pst
‘Mona didn’t wash the car but Jasper did.’
2006:(5))
Danish (Houser et al.
This talk evaluates two approaches to limited extraction from ellipsis sites.
1. The Derivational Approach:
Derivational approaches to ellipsis propose that licensing heads trigger ellipsis during the syntactic derivation. Ellipsis sites are frozen for syntactic operations.
§4 Phase-based ellipsis
Two approaches to phase-based ellipsis, and some of the issues they
raise.
§5
Conclusion
Final thoughts
2
Ellipsis is not spell-out
2
How to strand a verb
• Verbs are to the left of νP adverbs in Scandinavian: Verb-movement
For verb stranding verb-phrase ellipsis to occur, there are two requirements (Goldberg 2005).1
i. The language must have verb movement out of the verb phrase.
ii. The language must have vpe.
In this section, I briefly review verb movement and verb phrase ellipsis in a number
of relevant languages.
• I start with an overview of how we know verbs move in Scandinavian.
• I then turn to some relevant properties of vpe.
• Languages with vvpe have both vpe and verb-movement, so Scandinavian
poses a problem.
2.1 Verb Movement
There are a couple relevant ways to tell whether verbs have undergone movement.
• Position of the verb relative to adverbs/negation.
The destination of verb movement may vary from language to language.
• ν°:
English, Mainland Scandinavian embedded clauses
• T°:
Hebrew (Doron 1983), Irish (McCloskey 2011), Portuguese (Silva 2001)
• Asp°:
Russian (Bailyn 1995, Gribanova 2013)
• C°:
Germanic (matrix clauses) (Vikner 1995)
Additionally, I adopt the common assumption that the verb does not move to T° in
Scandinavian unless it also moves to C° (Vikner 1995, Westergaard 2009).2
• This means that only C° triggers movement of the verb out of νP.
• Position of subjects relative to the verb.
• There is no V°-to-T° movement.
This lets us determine where the verb sits on the surface.
• The assumption is that negation and certain adverbs sit between T° and ν°.
• When verbs appear to left of these adverbs or negation, verb movement must
have occurred. (Vikner 1995)
(4)
Peter (*drinks) [vP often (drinks) coffee].
Peter (drikker) [vP ofte (*drikker) kaffe].
Danish
Preview: The idea that verbs always move to C° in Scandinavian matrix clauses
is an important part of this analysis.
• The languages that have vvpe appear to have movement to a position
above ν° and below C°.
• For more on movement of the verb in Germanic, see Appendix A.
• Verbs are to the right of νP adverbs in English: No movement.
1
2
Other elements can be stranded – for example, auxiliaries. These do not clearly originate inside the ellipsis site. What makes verb-stranding remarkable is that an element originating inside the ellipsis site is
stranded, what Sailor (2014) calls X-stranding XP-ellipsis.
This is not a univerally held view; see Mikkelsen 2010, Thráinsson 1994, Travis 1984. It is also the case that some dialects appear to have movement to T°. However, there appear to be no cases of a dialect with
both V°-to-T° movement and vpe (Sailor 2014).
Nicholas LaCara
2.2 Verb phrase ellipsis
(8)
vpe targets a constituent roughly the size of a verb phrase.
I don’t know which puppy you will buy, but I know which onei you should
buy t i .
• Regularly strands auxiliaries and modals.3
• Adjuncts to verb phrases may also escape ellipsis.
• As in English, extraction out of ellipsis sites in Scandinavian languages appears
to be possible.
(5)
Ashley hasn’t gone to Tromsø, but Lindsay has νP .
(6)
Ele perguntou quem tinha comido o bolo, e ela perguntou quem
He asked
who had eaten the cake and she asked
who
não tinha νP .
not had
‘He asked who had eaten the cake, and she asked who had not eaten the
cake.’
a
Portuguese (Cyrino and Matos 2002:(22b))
(7)
Johan har inte läst Lolita, men Kalle har νP .
Johan has not read Lolita, but Kalle has
‘Johan has not read Lolita, but Kalle has read lolita.’
2012)
(9)
‘Which cakes do you want to bake, and which cakes don’t you?’ Norwegian (Bentzen et al. 2013)
(10)
Swedish (Thoms
This tells us the upper bound of the operation.
Hvilke kaker vil du bake, og [hvilke kaker]i vil du ikke bake t i ?
which cakes will you bake, and which cakes will you not bake
? Jeg ved ikke hvad for en hund Susan har valgt, men jeg ved [hvad
I know not what for a dog Susan has chosen but I know what
for en]k Marie har valgt tk .
for a/one Marie has chosen.
‘I don’t know what kind of dog Susah has picked, but I know what kind
Marie has.’6
Danish (L. Mikkelsen, p.c.)
• In general, νP appears to be the target (Aelbrecht 2010, Merchant 2013).4 This
is a point I will defend.
• The crucial ting here is that vpe targets the phrase in which the verb originates. We will only be concerned here with cases where verbs might escape
the ellipsis site.5
I assume that ellipsis sites contain unpronounced syntactic structure (Goldberg
2005, Merchant 2001, Schuyler 2001).
• This accounts for the fact that it is possible to extract material out of ellipsis
sites by movement:
• Scandinavian has an ellipsis process that targets verb phrases.
• It is generally possible to extract out of ellipsis sites, indicating that there
is internal structure to the site (cf. Houser et al. 2011).
3 It is possible to elide some auxiliaries (Lasnik 1995, Potsdam 1997, Warner 1985); in English, both auxiliary and passive be may be elided, whereas auxiliary have must usually be stranded.
4 There is purported variation. Some operations, like pseudogapping, might target larger chunks of material (Merchant 2013).
5 Verbs can be stranded under auxiliaries, though not always. Determining in which positions this is possible introduces a number of difficult complications, including which heads license ellipsis and where
the phase boundary is in the middle field, so I constrain the discussion to cases without auxiliaries.
6 Mikkelsen reports that her informant’s judgment came with the following comment: “I would give it a single question mark. I actually think I can do this, but I have to stop myself from adding a main verb. I
think it passes, but a little marked.”
3
4 Ellipsis is not spell-out
2.3 Summary: Putting it together
Two main assumptions:
We have seen that Goldberg’s conditions on vvpe are met in Scandinavian.
1. Ellipsis happens over the course of a derivation.
2. As soon as ellipsis occurs, the ellipsis site becomes frozen for further syntactic
operations, blocking further movement out of the ellipsis sites.
i. Scandinavian languages have verb movement out of the verb phrase.
ii. Scandinavian languages have vpe out of which material may be extracted.
7
Despite this, vvpe does not occur in Scandinavian. Do-support occurs instead.
(11)
a.
b.
Mona og Jasper vaskede bilen, eller rettere Mona gjorde νP .
Mona and Jasper wash.pst car.def, or rather Mona do.pst
‘Mona and Jasper washed the car, or rather Mona did.’
a
Danish (Houser et al. 2011:(10))
Johan leste ikke Lolita, men Marie gjorde.
Johan read not Lolita, but Marie did.
‘Johan didn’t read Lolita, but Marie did.’
a
Norwegian (Sailor 2014:(8b))
The question now is: Why?
• In the next two sections I look at theories of ellipsis that are designed in
part to capture the limited ability to extract out of some ellipsis sites.
Sailor (2014) argues that Aelbrecht’s (2010) approach accounts for the lack of vvpe
in Scandinavian straightforwardly.
• Verbs become frozen in ellipsis sites before they can move to C°.
• Languages that have V°-to-T° movement will still display vvpe.
In this section, I will demonstrate that the Aelbrecht/Sailor approach predicts the
correct distribution of vvpe across languages based on independent facts about head
movement in those languages.
3.1 Extraction from ellipsis sites
Let us assume a slightly simplified version of Aelbrecht’s (2010) approach to ellipsis
licensing.
1. When a licensing head L° merges, the constituent XP targeted for deletion is
frozen immediately. No further syntactic operations are possible.
2. Any material that moves to a position outside XP before or when L° merges
will escape ellipsis and be free for further operations;8 any material inside XP
will be elided and will not be available for further syntactic operations.
3
Derivational ellipsis
In light of the fact that not all extractions out of ellipsis sites are possible, derivational
accounts of ellipsis posit that ellipsis freezes ellipsis sites (Aelbrecht 2010, Baltin
2011).
7
• Thus, when T° merges, the material in νP will delete immediately.
• Since the verb remains in νP in English, there is no vvpe.
Platzack (2012) reports that ellipsis with göra in Swedish is ungrammatical, unlike in Danish and Norwegian. Instead, speakers must use the verbal pro-form det:
(i)
8
The typical assumption here is that T° licenses vpe.
Maria körde inte bilen, men Johan gjorde *(det).
Maria drove not car.def but Johan did it
‘Maria didn’t drive the car, but Johan did.’
Why this is the case is mysterious, but it remains the case that vvpe is still predicted to occur in Swedish when it does not. For more on det, see Houser et al. 2007, 2011 and Bentzen et al. 2013.
This is under the assumption that all operations triggered by a head happen simultaneously, allowing ellipsis and head movement to happen at the same time (Aelbrecht 2010:109, n.23).
Nicholas LaCara
(12)
John didn’t put the books on the table, but Mary did.
a.
b.
Step 1 – Build νP:
[νP Mary put+ν° [the books [put on the table]]]
Step 2 – Merge T°; Attract subject; Elide νP:
[TP Mary [T° [νP Mary put+ν° [the books [put on the table] ] ] ] ]
There’s nothing in T° here, so do-insertion must occur to fill T°.
• T° licenses ellipsis of νP before C° merges, so the main verb will be frozen in
the ellipsis site.
• Do-support rescues the material moved to C° (Platzack 2012).
Mona vaskede ikke bilen men Jasper gjorde νP .
Mona wash.pst not car.def, but Jasper do.pst
‘Mona didn’t wash the car but Jasper did.’
(14)
a.
3.2 Verb stranding vpe
Verb stranding may occur in languages with independent movement to T° or Asp°
on the assumption that T° licenses vpe.
• Assume that head movement happens in Syntax (Hartman 2011, Matushansky
2006, Roberts 2010).
• T° or Asp° triggers verb movement: Heads will move to those positions when
those heads merge.
• T° licenses ellipsis of νP, so it will attract the verb just in time.
(13)
Quando a Ana pôs
os óculos na
mesa, a Maria também
When the Ana put.pst the glasses on.the table, the Maria too
pôs.
put.pst
‘When Ana put the glasses on the table, the Maria did too.’
a.
Step 1 – Build νP:
[νP pôr+ν° [os óculos [pôr na mesa]]]
b.
Step 2 – Merge T°; Attract subject,verb; delete νP:
[TP a Maria também pôr+ν°+T° [νP a Maria pôr+ν° [os óculos
[pôr na mesa]]]]
3.3 Scandinavian
Verb stranding in vpe contexts in mainland Scandinavian fails because the verb is
attracted to C° after vpe is triggered.
• One extra step: C° merges, triggering verb movement in Scandinavian.
b.
c.
Danish
Step 1 – Build νP:
[νP Jasper vaske+ν° [vaske bilen]]
Step 2 – Merge T°; Attract subject; Delete νP:
[TP Jasper T0[pst] [νP Jasper vaske+ν° [vaske bilen] ] ]
Step 3 – Merge C°; Attract subject, T°:
[CP Jasper T°+C° [TP Jasper T° [νP Jasper vaske+ν° [vaske bilen] ] ] ]
Of interest here: At Step 2, Scandinavian and English look exactly the same.
• The difference between them reduces to Scandinavian being a V2 language
that requires a filled C°, but this is irrelevant to ellipsis.
3.4 Summary
This approach relies on two ideas:
• A derivational freezing account of ellipsis:
Material in an ellipsis site is frozen for further syntactic operations during the
course of the syntactic derivation.
• Syntactic head movement:
Head movement is triggered in the syntax, and the different triggers result in
timing differences.
Important fact: The derivations proceed identically.
• The derivation is intrinsically ordered by the merger of syntactic heads.
• The differences between languages are the result of independently motivated facts about verb movement in those languages.
• We do not need to stipulate any further differences between languages.
5
6
Ellipsis is not spell-out
4
Phases
Another approach to blocking movement out of ellipsis sites is linking ellipsis to
phases and Spell out.
• The idea, following Holmberg (2001), is that an ellipsis site is coterminous
with a Phase or a Spell-Out domain.
• A lot of recent work on this approach (including Bošković 2014, Gengel 2007,
Harwood 2013, Rouveret 2012).
Indeed, the idea of freezing material for further syntactic operations bears considerable resemblance to Phasal spell out (Aelbrecht 2010, 2014).
4.1 Phase head complements (pce)
Some theories (Gengel 2007, Rouveret 2012) propose that the complement of a
phase head is a potential target for ellipsis and that ellipsis happens as soon as that
phase head merges.
• This follows Chomsky’s (2000) definition of the pic:
(15)
Phase Impenetrability Condition (pic):
(Chomsky 2000:108)
The Domain of phase head H° is not accessible to operations outside HP;
only H° and its edge are accessible to such operations.
Making the common assumption that ν° is the phase head, when ν° merges its complement is Spelled-Out immediately.
• Following Chomsky’s (2000, 2001) Phase Impenetrability Condition,
material that has been spelled out is no longer accessible for syntactic computation.
• Any material inside VP when ν° merges will be inaccessible for further operations, including movement.
• If ellipsis sites are phases, then the apparent frozenness of ellipsis sites under
Aelbrecht’s approach is derived.
• This predicts that movement out of ellipsis sites should behave exactly the
same as movement out of lower phases
Aelbrecht (2010) argues against this on the basis that not all predicted extraction
possibilities are attested.
• If ellipsis sites are identified with phases, then extraction possibilities out of
ellipsis sites and phases should be the same, all else being equal.
Scandinavian provides further evidence against this view.
• There are two different approaches:
1. Phase-head complement ellipsis (pce):
Only the complement of a phase head can be targeted for ellipsis.
2. Whole phase ellipsis (wpe):
The whole phase may be targeted for deletion.
• Scandinavian is problematic for both of these approaches. The ability for V2
to occur in non-elliptical cases complicates the picture for each of these and
requires additional stipulations to get the facts straight.
As we will see, V2 introduces complications for this.
• Additional stipulations blocking verb movement in ellipsis contexts are necessary to make it work.
4.1.1 Verb movement and phases
Under the view of the pic in (15), verbs must move through ν° to escape spell-out.
• Since it is the complement of ν° that is spelled-out, any verb that does not
move through this position will be frozen for further syntactic operations.
It is necessary for the verb to move to ν° for V2 order to be possible.
• If verbs remained in the complement of ν°, then they could never move to any
higher positions.
• It is a fairly common assumption that verbs must move to ν° anyway (Chomsky 2000, Kratzer 1996, Marantz 1997).
Nicholas LaCara
4.1.2 VVPE and overgeneration
(18)
Deletion of phase head complements predicts correctly that vvpe should be available in languages with V°-to-T° movement:
(16)
Quando a Ana pôs
os óculos na
mesa, a Maria também
When the Ana put.pst the glasses on.the table, the Maria too
pôs.
put.pst
‘When Ana put the glasses on the table, the Maria did too.’
a
Portuguese (Cyrino and Matos 2002:(14a))
a.
Step One – Build VP:
√
[VP os óculos [V′ pôr na mesa] ]
b.
Step Two – Build νP, move verb, spell out:
√
√
[νP a Maria [ν′ pôr+ν° [VP os óculos [V′ pôr na mesa] ] ] ]
c.
Step Three – Merge T°; move verb, subject:
√
√
[TP a Maria [T′ também pôr+ν°+T° [νP a Maria [ν′ pôr+ν° VP ] ]
]]
The problem is that it predicts incorrectly that both Scandinavian and English should
also have vvpe.
(17)
Mona vaskede ikke bilen men Jasper gjorde / *vaskede.
Mona wash.pst not car.def, but Jasper do.pst / wash.pst
‘Mona didn’t wash the car but Jasper did.’
2006:(5′ /5′′ ))
Ashleigh didn’t wash the car, but Lindsay did / *washed.
a.
Step One
√ – Build VP:
[VP [V′ wash the car] ]
b.
Step Two – Build
√ νP, move verb, spell
√ out:
[νP Lindsay [ν′ wash+ν° [VP [V′ wash the car] ] ] ]
c. * Step Three – Merge T°; move subject:
√
√
[TP Lindsay [T′ T° [νP Lindsay [ν′ wash+ν° [VP [V′ wash the car]
]]]]]
In both cases, the problem can be located at Step Two.
• V°-to-ν° movement predicts that the verb should remain available to further
syntactic operations because it escapes the spell out domain.
• Remember: This has to happen for V2 to be possible.
Nearly every account of pce encounters this difficulty (see, for instance, Bošković
2014:69–70, Gengel 2007:237–238, and Rouveret 2012:946–948).
• It is worth noting here that the issue of blocking head movement out of the
ellipsis site does not arise under the derivational account.
4.1.3 Blocking short movement
Danish (Houser et al.
a.
Step One
√ – Build VP:
[VP [V′ vaske bilen] ]
b.
Step Two – Build
spell out:
√ νP, move verb,√
[νP Jasper [ν′ vaske+ν° [VP [V′ vaske bilen] ] ] ]
c.
Step Three – Merge T°; move subject:
√
[TP Jasper [T′ T° [νP Jasper [ν′ vaske+ν° VP ] ] ] ]
d. * Step Four – Merge
√ C°; move verb, subject:
[√CP Jasper [C′ vaske+ν°+T° +C° [TP Jasper [T′ T° [νP Jasper [ν′
vaske+ν° VP ] ] ] ] ]
The question: Can anything else block verb movement in ellipsis contexts?
• In most cases, we need V°-to-ν° outside of, ellipsis contexts, so we must tie the
inability for verbs to move out of ellipsis sites to ellipsis specifically.
• There are two proposals I know of:
A) Do-support blocks verb movement.
B) The [e] feature blocks movement.
• Neither of these is particularly satisfying.
7
8
Ellipsis is not spell-out
Hypothesis A – Do-support blocks verb movement (Rouveret 2012).
• Claim: In English, there is a focus feature on T° associated with ellipsis that
must be lexicalized.
– Since there is no V°-to-T° movement in English, do must be merged in
ν° as an element which can move to T°.9
– Since do is in ν°, verbs cannot move through this position (Travis 1984).
Under this view, do-support cannot be a last-resort operation.
• Note that this is a loosening of the notion that ellipsis is coterminous with a
spell-out domain – we are no longer looking at just phase-head complements.
• This means that the ellipsis domain can be bigger than the typical spell-out
domain.
These approaches employ the definition of the pic introduced in Chomsky 2001. For
a phase heads H°, Z°,
(20)
• Do-support is not available in neutral contexts, so it cannot be freely merged
otherwise:
(19)
• This means that it is possible to extract material out of HP until Z° is merged.
* Johan gjorde öppnade dörren.
Johan did opened door.def
Intended: ‘Johan opened the door.’
(Revised) Phase Impenetrability Condition: (pic)
(Chomsky 2001:13)
The domain of H is not accessible to operations at ZP; only H and its edge
are accessible to such operations.
• Once Z° is merged, HP is frozen.
Swedish (Platzack 2012:(39))
The predictions this makes are slightly different from pce.
• This might work for English, but verbs probably move through T° in Scandinavian on their way to C°, so there is no explanation why they could not
lexicalize T°, as they do under normal circumstances.
Hypothesis B – [e] feature blocks head movement (Gengel 2007).
• The predictions this makes for Scandinavian will hinge in part on what until
means here.
• Gengel proposes that [e] sits on ν°. The presence of [e] blocks movement of
V°-to-ν°.
4.2.1 Potential overgeneration
• This is an ad hoc explanation for English.
wpe correctly predicts the availability of vvpe in V°-to-T° languages.
– While it would explain Scandinavian, too, there is no explanation for
why verbs can move through ν° in languages that vvpe.
4.2 Whole phases (wpe)
Some approaches to phase-based ellipsis propose that ellipsis targets not just the
phase-head complement, but the entire phase (Bošković 2014, Harwood 2013, Holmberg 2001).
9
• If νP is the target, rather than VP, then material in νP is not frozen until C°
merges.
• V°-to-T° predicts means that the verb escapes νP.
(21)
Quando a Ana pôs
os óculos na
mesa, a Maria também
When the Ana put.pst the glasses on.the table, the Maria too
pôs.
put.pst
‘When Ana put the glasses on the table, the Maria did too.’
a
Portuguese (Cyrino and Matos 2002:(14a))
Rouveret’s theory of ellipsis relies on the idea that inflectional morphology is available at the νP level, so do must be inserted in ν° for it to receive inflectional morphology.
Nicholas LaCara
a.
Step One – Build VP:
√
[VP os óculos [V′ pôr na mesa] ]
a.
b.
Step Two – Build νP, move verb:
√
√
[νP a Maria [ν′ pôr+ν° [VP os óculos [V′ pôr na mesa] ] ] ]
c.
Step Three – Merge T°; move verb, subject:
√
√
[TP a Maria [T′ também pôr+ν°+T° [νP a Maria [ν′ pôr+ν° [VP os
√
óculos [V′ pôr na mesa] ] ] ] ] ]
d.
Step Four – Merge C°; Spell out νP:
√
√
[CP C° [TP a Maria [T′ também pôr+ν°+T° [νP a Maria [ν′ pôr+ν°
√
[VP os óculos [V′ pôr na mesa] ] ] ] ] ] ]
Unlike the pce approach, wpe does not predict that English should have vvpe, since
the verb never leaves νP.
(22)
a.
Step One
√ – Build VP:
[VP [V′ wash the car] ]
b.
Step Two – Build
√ νP, move verb: √
[νP Lindsay [ν′ wash+ν° [VP [V′ wash the car] ] ] ]
d.
Step Three – Merge T°; move subject:
√
√
[TP Lindsay [T′ T° [νP Lindsay [ν′ wash+ν° [VP [V′ wash the car]
]]]]]
Step Four – Merge C°; Spell-out:
√
√
[CP C° [TP Lindsay [T′ T° [νP Lindsay [ν′ wash+ν° [VP [V′ wash
the car] ] ] ] ] ] ]
If we assume that the operations triggered by C° occur simultaneously (Chomsky
2001), as soon as it merges, wpe predicts that Scandinavian should have vvpe in V2
contexts.
(23)
Mona vaskede ikke bilen men Jasper gjorde / *vaskede νP .
Mona wash.pst not car.def, but Jasper do.pst / wash.pst
‘Mona didn’t wash the car but Jasper did.’
2006:(5′ /5′′ ))
The reason for this is that Spell-Out occurs too late in the derivation to block verb
movement out νP.
• Following the disccuion in §2, V2 is triggered by C° (Vikner 1995).
• We know that the verb escapes spell-out to make it to C° in non-elliptical contexts.
Ashleigh didn’t wash the car, but Lindsay did.
c.
Step One
√ – Build VP:
[VP [V′ vaske bilen] ]
b.
Step Two – Build
√ νP, move verb:√
′
[νP Jasper [ν vaske+ν° [VP [V′ vaske bilen] ] ] ]
c.
Step Three – Merge T°; move subject:
√
√
[TP Jasper [T′ T° [νP Jasper [ν′ vaske+ν° [VP [V′ vaske bilen] ] ] ]
]]
d. * Step Four – Merge
√ C°; move verb, subject; Spell out νP:
[√CP Jasper [C′ vaske+ν°+T°+C°
[TP Jasper [T′ T° [νP Jasper [ν′
√
vaske+ν° [VP [V′ vaske bilen] ] ] ] ] ] ]
Danish (Houser et al.
• So it follows that if ellipsis is a reflex of spell-out, verbs should be able to move
to C° out of ellipsis sites.
In other words, we know that verbs can escape spell-out in non-elliptical contexts,
so again, it must be something about ellipsis that blocks movement of the verb to C°.
4.2.2 Ordering and extraction
One way of dealing with this is to try to order the operations triggered by a head –
i.e., posit that operations do not happen simultaneously.
• Bošković (2014:46) claims that freezing of material in the Phase happens before C° can attract any material.
• If the whole phase is frozen before head movement can occur, then vvpe will
be blocked.
However, if ellipsis targets the whole phase, this means that the phase edge will be
frozen.
• This means that even A- and A′ -extraction will be blocked out of ellipsis sites
when wpe is used.
9
10 Ellipsis is not spell-out
But extraction is possible out of ellipsis sites, as we saw in Section 2.2.
Overall, Scandinavian poses problems for phase-based accounts.
• In pce accounts, we must introduce stipulations to block movement out of
ellipsis sites – the freezing effects from the pic are not enough to do this.
• In fact, Bošković claims that ellipsis of the phase head complement – basically,
pce– must be available for wh-extraction.
• The wpe approach can be made to work if we adopt some of Bošković’s positions, but the status of these remains unclear.
Since wh-extraction is available in Scandinavian, this means that pce must be available.
• We already know that pce makes the wrong predictions for Scandinavian,
though.
• If pce is available for wh-extraction, it should also be available for verb movement.
Ultimately, ordering operations triggered by C° requires pce to be available, but pce
predicts Scandinavian should have vvpe.
4.3 Summary
pce accounts can only block movement of the verb from ellipsis sites with additional
stipulations.
• V°-to-ν° movement must be blocked by some mechanism, but this is generally
hard to motivate.
wpe approaches do better.
• However, we must widen the possible targets of vpe to be slightly bigger than
typical Spell-out domains, a step away from the original insight.
• We also need to stipulate the ordering of operations triggered by a phase head.
• This also requires pce when extraction is possible. It remains unclear if vpe
has different targets in different cases.
The derivational account does not need to do any of this.
• By positing that T° licenses ellipsis rather than C°, the distribution of vvpe is
readily accounted for.
• νP is always the target.
5
Conclusion
The derivational approach to ellipsis requires us to make fewer assumptions than
does the phase-based approach.
• Out of the box, it correctly predicts the distribution of vvpe.
• Languages where verb movement is triggered after T° merges should not have
vvpe.
To the extent that this account is successful, it carries other implications for the grammar as well.
• Head movement
Chomsky (2001) suggests that head movement may be relegated to PF. However, the interaction of derivational ellipsis with head movement in the narrow
syntax properly predicts the distribution of vvpe. PF accounts of head movement must stipulate the order of head movement relative to ellipsis (Boeckx
and Stjepanović 2001).
• Do-support in Scandinavian
Lots of recent work has gone into do-support in Scandinavian. Houser et al.
(2011) argue that Danish gøre, ‘do’, is a special element necessary to license
empty νPs. This analysis brings ellipsis in Scandinavian in line with other languages and is more compatible with Platzack’s (2012) last-resort analysis of
do-support in Scandinavian.
• Destination of verb movement in Germanic
There has always been debate about whether verbs move to T° independently
of movement to C° in Germanic matrix clauses (Mikkelsen 2010, Thráinsson
1994, Travis 1984). The arguments in this paper rest on the idea that verbs
move to C° in mainland Scandinavian matrix clauses and not to T°. This can
be seen as support for this view that verbs always move to C°.
Nicholas LaCara 11
A
Verb second and verb movement
In Mainland Scandinavian, placement of the verb is different depending on whether
the clause is embedded.
• In matrix clauses (24), verbs show up to the left of negation and VP-adverbs.
• In embedded clauses (25), verbs (by default) show up to the right of negation
and VP-adverbs.
(24)
• SpecCP is first position.
• C° is second position.
• When not in first position, subjects are in SpecTP since they are to the left of
νP adverbs.
(27)
a.
C
at DP
drikker
T
vP
DP
T
vP
Peter
ofte
ofte
vP
vP
DP drikker kaffe
DP V° kaffe
In matrix clauses, the finite verb is always preceded by some phrasal element, typically the subject as in (24) above
We know verbs do not move to T° independent of movement to C° through comparison to Icelandic.
• The first element can be some other fronted element, like the PP om morgonen
in (26).
• In Icelandic, verbs always come to the left of adverbs, even in embedded
clauses (28).
• In these cases the subject remains in SpecTP, providing evidence that the verb
has moved out of TP.
• In Danish embedded clauses, verbs come to the right of adverbs (29)
1
(26)
2
z
}|
{ z }| {
Om morgonen drikker Peter ofte kaffe.
in morning.def drinks Peter often coffee.
(28)
• The position of the verb is traditionally called second position.
• The the element before the verb is said to be in first position.
The standard account: In matrix clauses there is V°-to-C° movement, whereas verbs
in embedded clauses remain in situ (den Besten 1983, Vikner 1995).10
Icelandic – V°-to-T°:
a.
‘In the morning Peter often drinks coffee.’ Danish (Vikner 1995:47, (33e))
10
TP
TP
Peter
Vi ved [CP at Peter ofte drikker kaffe om morgenen]
we know [ that Peter often drinks coffee in morning.def]
‘We know that Peter often drinks coffee in the morning’
a
Danish (Vikner 1995:47, (33f))
CP
C
DP
Peter drikker ofte kaffe om morgenen.
Peter drinks often coffee in morning.def
‘Peter often drinks coffee in the morning.’ Danish (Vikner 1995:47, (33c))
(25)
b.
CP
(Vikner 1995:145)
Ég spurði af hverju Pétur hafði oft lesið hana.
I asked why
Peter had often read it.
b. * Ég spurði af hverju Pétur oft hafði lesið hana.
I asked why
Peter often had read it.
(29)
Danish – V° in situ:
a. * Jeg spurgte hvorfor Peter havde ofte læst den.
I asked why Peter had
often read it.
b.
Jeg spurgte hvorfor Peter ofte havde læst den.
Peter often had
read it.
Some authors have argued that the verb does not always make it to C° in matrix clauses (see, for instance, Mikkelsen 2010). I asked why
(Vikner 1995:145)
12 Ellipsis is not spell-out
This is taken as evidence that there is V°-to-T° movement in Icelandic, but not in
Mainland Scandinavian.11
In Mainland Scandinavian:
• Verbs move to the left periphery in matrix clauses.
• They remain in situ in embedded clauses.
• There is no independent movement to T°.
Acknowledgments
I am indebted to Lobke Aelbrecht, Matt Barros, Kyle Johnson, Stefan Keine, Fernanda Mendes, Jon Ander Mendia, Line Mikkelsen, Ethan Poole, Alain Rouveret,
Craig Sailor, Ellen Woolford, Luis Vicente, participants in the Syntax Seminar at
UMass (Fall 2012), and the audience at the 29th Comparative Germanic Syntax
Workshop at the University of York, 2014. Many thanks are also due to the anonymous cgsw reviewer who made me aware of Craig Sailor’s convergent work on ellipsis in Scandinavian, and to the three nels reviewers who provided relevant criticism for an abstract of this paper. Parts of this research were supported by summer
graduate student research grants from the UMass Department of Linguistics.
References
Aelbrecht, Lobke. 2010.
The Syntactic Licensing of Ellipsis.
Linguistik
Actuell/Linguistics Today. John Benjamins.
Aelbrecht, Lobke. 2014. What ellipsis can do for phases, and what it can’t, but not
why. Ms. University of Ghent. Under review for Special issue of The Linguistic
Review.
Bailyn, John. 1995. A Configurational Account of Russian ”Free” Word Order. Doctoral Dissertation, Cornell University.
Baltin, Mark. 2011. Deletion Versus Pro-Forms: An Overly Simple Dichotomy?
Natural Language and Linguistic Theory URL http://www.springerlink.com/
content/7171160x0305h4l5, published online, November 2011.
Bentzen, Kristine, Jason Merchant, and Peter Svenonius. 2013. Deep properties
of surface pronouns: Pronominal predicate anaphors in Norwegian and German.
The Journal of Comparative Germanic Linguistics 16:97–125.
den Besten, Hans. 1983. On the the Interaction of Root Transformations and Lexical
Deletive Rules. In On the Formal Syntax of the Westgermania, ed. W. Abrahama,
47–131. Amsterdam: John Benjamins.
Boeckx, Cedric, and Sandra Stjepanović. 2001. Head-ing toward PF. Linguistic Inquiry 32:345–355.
Bošković, Željko. 2014. Now I’m a Phase, Now I’m Not a Phase: On the Variability
of Phases with Extraction and Ellipsis. Linguistic Inquiry 45:27–89.
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Step by Step, ed.
R. Martin et al., 89–155. Cambridge, Mass: MIT Press.
Chomsky, Noam. 2001. Derivation by Phase. In Ken Hale: A Life in Language, ed.
Michael Kenstowicz, 1–52. Cambridge, Mass: MIT Press.
Cyrino, Sonia M. L., and Gabriela Matos. 2002. VP ellipsis in European and Brazilian
Portuguese – a comparative analysis. Journal of Portuguese Linguistics 1:177–195.
Doron, Edit. 1983. Verbless Predicates in Hebrew. Doctoral Dissertation, Univeristy
of Texas at Austin.
Doron, Edit. 1990. V Movement and VP-Ellipsis. Ms., The Hebrew University of
Jerusalem. Published as Doron 1999.
Doron, Edit. 1999. V Movement and VP-Ellipsis. In Fragments: Studies in Ellipsis and
Gapping, ed. S. Lappin and E. Benmamoun, 124–140. Oxford: Oxford University
Press.
Gengel, Kirsten. 2007. Phases and Ellipsis. In Proceedings of the thirty-seventh annual
meeting of the North East Linguistic Society, ed. Emily Elfner and Martin Walkow.
Amherst, Mass.: GLSA Publications.
Goldberg, Lotus. 2005. Verb-Stranding VP Ellipsis: A Cross-Linguistic Study. Doctoral Dissertation, McGill, Monteal, QC. URL http://www.lotusgoldberg.net/
dissertation.html.
Gribanova, Vera. 2013. Verb-stranding verb phrase ellipsis and the structure of the
Russian verbal complex. Natural Language and Linguistic Theory 31:91–136.
Hartman, Jeremy. 2011. The Semantic Uniformity of Traces: Evidence from Ellipsis
Parallelism. Linguistic Inquiry 42:367–388.
11 A quick aside: There is no vpt or vpe in Icelandic (Platzack 2012), so we cannot compare the way it behaves to Mainland Scandinavian. Many languages lack one operation or the other; for example, German
lacks vpe (Lobeck 1995). It is unclear why this should be. Many authors have attempted to answer this question (e.g. Doron 1990, Merchant 2001, Rouveret 2012), but it has been very hard to pin down a
property that accounts for which languages will have vpe and which will not. This is a very large question that I sadly have little to say about.
Nicholas LaCara 13
Harwood, William. 2013. Being Progressive is Just a Phase: Dividing the Functional
Hierarchy. Doctoral Dissertation, University of Ghent.
Holmberg, Anders. 2001. The syntax of yes and no in Finnish. Studia Linguistica
55:141–175.
Houser, Michael, Line Mikkelsen, Ange Strom-Weber, and Maziar Toosarvandani.
2006. Gøre-Support in Danish. Talk given at the 21st Comparative Germanic
Syntax Workshop, 1 April 2006.
Houser, Michael J., Line Mikkelsen, and Maziar Toosarvandani. 2007. Verb Phrase
Pronominalization in Danish: Deep or Surface Anaphora? In Proceedings of the
Thirty-Fourth Western Conference on Linguistics, ed. Erin Brainbridge and Brian
Agbayani, 183–195.
Houser, Michael J., Line Mikkelsen, and Maziar Toosarvandani. 2011. A Defective
Auxiliary in Danish. Journal of Germanic Linguistics 23:245–298.
Kratzer, Angelika. 1996. Severing the External Argument from its Verb. In Phrase
Structure and the Lexicon, ed. Johan Rooryck and Laurie Zaring, Studies in Natural Language & Linguistic Theory, 109–137. Dordrecht: Kluwer Academic Publishers.
Lasnik, Howard. 1995. Verbal Morphology: Syntactic Structures meets the Minimalist
Program. In Evolution and Revolution in Linguistic Theory: Essays in Honor of Carlos Otero, ed. Paula Kempchinsky and Héctor Campos. Georgetown University
Press.
Lipták, Anikó, and Andrés Saab. 2014. No N-raising out of NPs in Spanish: ellipsis
as a diagnostic of head movement. Natural Language and Linguistic Theory .
Lobeck, Anne. 1995. Ellipsis. New York: Oxford University Press.
Marantz, Alec. 1997. No Escape From Syntax: Don’t Try Morphological Analysis in
the Privacy of Your Own Lexicon. In Proceedings of the 21st Annual Penn Linguistics
Colloquium, ed. Alexis Dimitriadis, volume 4 of Penn Working Papers in Linguistics,
201–225.
Matushansky, Ora. 2006. Head Movement in Linguistic Theory. Linguistic Inquiry
37:69–109.
McCloskey, James. 2011. The Shape of Irish Clauses. In Formal Approaches to Celtic
Linguistics, ed. Andrew Carnie, 143–178. Newcastle upon Tyne: Cambridge Scholars Publishing.
Merchant, Jason. 2001. The Syntax of Silence: Sluicing, Islands, and the Theory of Ellipsis. Oxford Studies in Theoretical Linguistics. Oxford University Press.
Merchant, Jason. 2013. Voice and ellipsis. Linguistic Inquiry 44:77–108.
Mikkelsen, Line. 2010. On what comes first in a verb-second language. Ms. Under
review.
Platzack, Christer. 2012. Cross Germanic variation in the realm of support verbs. In
Comparative Germanic syntax: the state of the art, ed. Peter Ackema, Rhona Alcorn,
Caroline Heycock, Dany Jaspers, Jeroen Van Craenenbroeck, and Guido van den
Wyngaerd, 279–309. Amsterdam: John Benjamins.
Potsdam, Eric. 1997. English Verbal Morphology and VP ellipsis. In The Proceedings
of the 27th Meeting of the North East Linguistic Society, 353–368.
Roberts, Ian. 2010. Agreement and Head Movement: Clitics, Incorporation, and Defective Goals. Linguistic Inquiry Monographs. Cambridge, Mass.: MIT Press.
Rouveret, Alain. 2012. VP ellipsis, phases and the syntax of morphology. Natural
Language and Linguistic Theory 30:897–963.
Sailor, Craig. 2014. The Typology of Head Movement and Ellipsis: A reply to Lipták
& Saab. Ms. University of Groningen. lingbuzz/002148. Accepted for publication
in Natural Language & Linguistic Theory.
Schuyler, Tami. 2001. Wh-Movement out of the Site of VP Ellipsis. Master’s thesis,
University of California, Santa Cruz.
Silva, Gláucia V. 2001. Word order in Brazilian Portuguese, volume 57 of Studies in
Generative Grammar. New York: Mouton de Gruyter.
Thoms, Gary. 2012. Ellipsis licensing and the variety of verb movement in Germanic.
Talk given at the 27th Comparative Germanic Syntax Workshop, Yale University,
31 May 2012.
Thráinsson, Höskuldur. 1994. Comments on the paper by Vikner. In Verb Movement,
ed. David Lightfoot and Norbert Hornstein. Cambridge: Cambridge University
Press.
Travis, Lisa DeMena. 1984. Parameters and Effects of Word Order Variation. Doctoral Dissertation, Massachusetts Institute of Technology, Cambridge, MA.
Vikner, Sten. 1995. Verb Movement and Expletive Subjects in the Germanic Languages.
Oxford Studies in Comparative Syntax. New York: Oxford University Press.
Warner, Anthony R. 1985. The Structure of English Auxiliaries: A Phrase Structure
Grammar. Bloomington, Indiana: Indiana University Linguistics Club.
Westergaard, Marit. 2009. The Acquisition of Word Order. Amsterdam: John Benjamins.